Study On Increased Fusion Map Based On RGB-D Sensor

With the development of the artificial intelligence theory, computer and sensor technology, the study on robot has been carried out widely at home and abroad. In recent years, the mobile robot has more and more applications in industry and social service and so on. SLAM(Simultaneous Localization and Mapping) is the core of the mobile robot research subject. It has important theoretical significance and application value.This paper mainly studies on how to create the fusion map in unknown environment. In order to create the fusion map fast and accurately, this paper puts forward a kind of method to create increased fusion map combining topological and metric map in unknown environment based on RGB-D sensor. The method describes environment at different levels by combining the topological map, the local metric map and the global metric map. Using strong calculation ability of topological map to provide environment connectivity in global environment and support the fusion map. At the same time, metric map describes the Irregular obstacle in the environment in detail. Fusion map is a kind of environment map that can describe outline details on the premise of guaranteeing the algorithm computing ability. When creating the topology map, the method to select topology point is proposed based on VFH, it reduces the dimensions of the 3D space information to VFH firstly, then choose topology point according to the Euclidean distance of VFH. In the topology point connection, this paper proposes topology point connection mode based on metric map and greedy algorithm, converts topology point connection to the path planning, and establishes the topology network successfully. When creating global metric map, the cumulative error is always the key point of the research. In order to solve the problem of cumulative error, this paper presents image closed loop algorithm based on layered topology point to fix map deviation caused by sensor errors. Topological and metric map rely on each other, environment model is built on the basis of the robot localization finally. At last, the fusion map combining topological and metric map is built through the experiment successfully.In this paper, we use visual measurement technology, provide a method to create increased fusion map based on RGB-D sensor, provide theoretical basis and technical support for robot to realize autonomous operation function, make the mobile robot a wide range of applications in unknown and complex environment, can promote the development of subjects and fields related to robot technology.